Optimization of Freeform Building Shape Using Genetic Algorithm

Abstract

This paper proposes an optimization algorithm for envelope design during an early stage freeform building design. This study performs geometric modeling using a three-dimensional (3D) design tool called Rhino, which designs examples of freeform buildings with a parametric design method, i.e., a 3D design technique for freeform building design. This method proceeds by mesh generation for the freeform building base using finite element analysis (FEM; a finite design method). The design variables from the thermal load prediction model are transformed based on the geometric information. This study used a genetic algorithm (GA) to minimize the envelope thermal load of freeform buildings. The results of the optimization process, which were applied to freeform building, indicated a 23.6–52.3% discrepancy between the initial and the optimized forms. We found that the optimized form consumed 23.6–52.3% less energy than the initial buildings produced by the parametric design method, which was facilitated by tuning each parameter for energy-conserving freeform buildings.